The sonogram is currently an effective cancer screening and diagnosis way due to the convenience and harmlessness in humans. Traditionally, lesion boundary segmentation is first adopted and then classification is conducted, to reach the judgment of benign or malignant tumor. In addition, sonograms often contain much speckle noise and intensity inhomogeneity. This study proposes a novel benign or malignant tumor classification system, which comprises intensity inhomogeneity correction and stacked denoising autoencoder (SDAE), and it is suitable for small-size dataset. A classifier is established by extracting features in the multilayer training of SDAE; automatic analysis of imaging features by the deep learning algorithm is applied on image classification, thus allowing the system to have high efficiency and robust distinguishing. In this study, two kinds of dataset (private data and public data) are used for deep learning models training. For each dataset, two groups of test images are compared: the original images and the images after intensity inhomogeneity correction, respectively. The results show that when deep learning algorithm is applied on the sonograms after intensity inhomogeneity correction, there is a significant increase of the tumor distinguishing accuracy. This study demonstrated that it is important to use preprocessing to highlight the image features and further give these features for deep learning models. In this way, the classification accuracy will be better to just use the original images for deep learning.

Background: The number of breast cancer patients has increased each year, and the demand for breast cancer detection has become quite large. There are many common breast cancer diagnostic tools. The latest automated whole breast ultrasound (ABUS) technology can obtain a complete breast tissue structure, which improves breast cancer detection technology. However, due to the large amount of ABUS image data, manual interpretation is time-consuming and labor-intensive. If there are lesions in multiple images, there may be some omissions. In addition, if further volume information or the three-dimensional shape of the lesion is needed for therapy, it is necessary to manually segment each lesion, which is inefficient for diagnosis. Therefore, automatic lesion segmentation for ABUS is an important issue for guiding therapy. Methods: Due to the amount of speckle noise in an ultrasonic image and the low contrast of the lesion boundary, it is quite difficult to automatically segment the lesion. To address the above challenges, this study proposes an automated lesion segmentation algorithm. The architecture of the proposed algorithm can be divided into four parts: (I) volume of interest selection, (II) preprocessing, (III) segmentation, and (IV) visualization. A volume of interest (VOI) is automatically selected first via a three-dimensional level-set, and then the method uses anisotropic diffusion to address the speckled noise and intensity inhomogeneity correction to eliminate shadowing artifacts before the adaptive distance regularization level set method (DRLSE) conducts segmentation. Finally, the two-dimensional segmented images are reconstructed for visualization in the three-dimensional space. Results: The ground truth is delineated by two radiologists with more than 10 years of experience in breast sonography. In this study, three performance assessments are carried out to evaluate the effectiveness of the proposed algorithm. The first assessment is the similarity measurement. The second assessment is the comparison of the results of the proposed algorithm and the Chan-Vese level set method. The third assessment is the volume estimation of phantom cases. In this study, in the 2D validation of the first assessment, the area Dice similarity coefficients of the real cases named cases A, real cases B and phantoms are 0.84±0.02, 0.86±0.03 and 0.92±0.02, respectively. The overlap fraction (OF) and overlap value (OV) of the real cases A are 0.84±0.06 and 0.78±0.04, real case B are 0.91±0.04 and 0.82±0.05, respectively. The overlap fraction (OF) and overlap value (OV) of the phantoms are 0.95±0.02 and 0.92±0.03, respectively. In the 3D validation, the volume Dice similarity coefficients of the real cases A, real cases B and phantoms are 0.85±0.02, 0.89±0.04 and 0.94±0.02, respectively. The overlap fraction (OF) and overlap value (OV) of the real cases A are 0.82±0.06 and 0.79±0.04, real cases B are 0.92±0.04 and 0.85±0.07, respectively. The overlap fraction (OF) and overlap value (OV) of the phantoms are 0.95±0.01 and 0.93±0.04, respectively. Therefore, the proposed algorithm is highly reliable in most cases. In the second assessment, compared with Chan-Vese level set method, the Dice of the proposed algorithm in real cases A, real cases B and phantoms are 0.84±0.02, 0.86±0.03 and 0.92±0.02, respectively. The Dice of Chan-Vese level set in real cases A, real cases B and phantoms are 0.65±0.23, 0.69±0.14 and 0.76±0.14, respectively. The Dice performance of different methods on segmentation shows a highly significant impact (P<0.01). The results show that the proposed algorithm is more accurate than Chan-Vese level set method. In the third assessment, the Spearman’s correlation coefficient between the segmented volumes and the corresponding ground truth volumes is ρ=0.929 (P=0.01). Conclusions: In summary, the proposed method can batch process ABUS images, segment lesions, calculate their volumes and visualize lesions to facilitate observation by radiologists and physicians.

In the registration of medical images, nonrigid registration targets, images with large displacement caused by different postures of the human body, and frequent variations in image intensity due to physiological phenomena are substantial problems that make medical images less suitable for intensity-based image registration modes. These problems also greatly increase the difficulty and complexity of feature detection and matching for feature-based image registration modes. This research introduces an automatic image registration algorithm for infrared medical images that offers the following benefits: effective detection of feature points in flat regions (cold patterns) that appear due to changes in the human body’s thermal patterns, improved mismatch removal through coherent spatial mapping for improved feature point matching, and large-displacement optical flow for optimal transformation. This method was compared with various classical gold standard image registration methods to evaluate its performance. The models were compared for the three key steps of the registration process—feature detection, feature point matching, and image transformation—and the results are presented visually and quantitatively. The results demonstrate that the proposed method outperforms existing methods in all tasks, including in terms of the features detected, uniformity of feature points, matching accuracy, and control point sparsity, and achieves optimal image transformation. The performance of the proposed method with four common image types was also evaluated, and the results verify that the proposed method has a high degree of stability and can effectively register medical images under a variety of conditions.

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is used to diagnose breast disease. Obtaining anatomical information from DCE-MRI requires the skin be manually removed so that blood vessels and tumors can be clearly observed by physicians and radiologists; this requires considerable manpower and time. We develop an automated skin segmentation algorithm where the surface skin is removed rapidly and correctly. The rough skin area is segmented by the active contour model, and analyzed in segments according to the continuity of the skin thickness for accuracy. Blood vessels and mammary glands are retained, which remedies the defect of removing some blood vessels in active contours. After three-dimensional imaging, the DCE-MRIs without the skin can be used to see internal anatomical information for clinical applications. The research showed the Dice’s coefficients of the 3D reconstructed images using the proposed algorithm and the active contour model for removing skins are 93.2% and 61.4%, respectively. The time performance of segmenting skins automatically is about 165 times faster than manually. The texture information of the tumors position with/without the skin is compared by the paired t-test yielded all p < 0.05, which suggested the proposed algorithm may enhance observability of tumors at the significance level of 0.05.